Generall overhead in XHR response?! That makes no sense...

[stevoffm]

Hi,

I am just evaluating some jsf-ajax frameworks and profiled some components of richfaces live demo and also my own richfaces sample application with firebug. I found out that all performed XHR have the complete page content in the response part, which creates much uneeded overhead.

For example the simple Ajax Echo in the UserGuide:

<h:inputText size="50" value="#{ajaxBean.text}">
 <a4j:support event="onkeyup" reRender="rep" />
 </h:inputText>
<h:outputText value="#{ajaxBean.text}" id="rep" />

Produces a 2KB XHR each time a key is pressed. For example the
Apache Trinidad Framework only needs 800 Bytes for the same operation.
Thats because not the whole page is sent but only the necessary data.

I already tried it with surrounding a4j:region tags with renderRegionOnly="true", but still the same overhead.

My question: Is thats a generall richfaces design issue or is it just a configuration and setup problem?? Can the size be reduced, by maybe change the format of the message to XML and only transfer essential data?
In my oppinion this is the concept behind ajax! Why make an XHR when nevertheless the whole page is transferred anyway?

Any answers are welcome!!

Cheers and Thanks!


BTW.: these two threads concern the same issue, but I did not get a real answer out of them:

http://jboss.com/index.html?module=bb&op=viewtopic&t=126882
http://jboss.com/index.html?module=bb&op=viewtopic&t=121695

[sergeysmirnov]

Does Trinidad has a4j:support (or whatever:support) ?

[sergeysmirnov]

"StevoFFM" wrote:
Thats because not the whole page is sent but only the necessary data.

I guess, this phrase means that RichFaces sends the whole page back on each Ajax response. If so, you made your exploring not too accurate.

[stevoffm]

No, Trinidad has costum components, where you can enable ajax if you want.
For example:

<tr:form id="myForm">
<tr:inputText value="#{ajaxBean.text}" autoSubmit="true"
id="myInput" />
<tr:outputText value="#{ajaxBean.text}" partialTrigers=
"myInput" id="myOutput" />
</tr:form>

partialTrigers and autoSubmit attributes enable ajax functionality in this case.

[sergeysmirnov]

FireBug report on first request at:
http://livedemo.exadel.com/richfaces-demo/richfaces/support.jsf

Request Headers
Host livedemo.exadel.com
User-Agent Mozilla/5.0 (Windows; U; Windows NT 5.2; en-US; rv:1.8.1.9) Gecko/20071025 Firefox/2.0.0.9
Accept text/xml,application/xml,application/xhtml+xml,text/html;q=0.9,text/plain;q=0.8,image/png,*/*;q=0.5
Accept-Language en-us,en;q=0.5
Accept-Encoding gzip,deflate
Accept-Charset ISO-8859-1,utf-8;q=0.7,*;q=0.7
Keep-Alive 300
Proxy-Connection keep-alive
Content-Type application/x-www-form-urlencoded; charset=UTF-8
Referer http://livedemo.exadel.com/richfaces-demo/richfaces/support.jsf;jsessionid=CBD3B0656620E9A7A1AA3E5DCECAB2F4?c=support
Content-Length 125
Cookie JSESSIONID=CBD3B0656620E9A7A1AA3E5DCECAB2F4; __utmz=141944294.1201207494.8.3.utmccn=(referral)|utmcsr=jsftutorials.net|utmcct=/helpDesk/|utmcmd=referral; __utma=141944294.723785644.1197082332.1200374661.1201207494.8
Pragma no-cache
Cache-Control no-cache

[stevoffm]

I guess, this phrase means that RichFaces sends the whole page back on each Ajax response. If so, you made your exploring not too accurate.

Okay where is my mistake? As a matter of fact the "ajax support" example in live demo produces a 78KB big XHR with each key stroke, conatinig the whole page in the response.

Can this be avoided?

[stevoffm]

Request Headers
Host livedemo.exadel.com
User-Agent Mozilla/5.0 (Windows; U; Windows NT 5.2; en-US; rv:1.8.1.9) Gecko/20071025 Firefox/2.0.0.9
Accept text/xml,application/xml,application/xhtml+xml,text/html;q=0.9,text/plain;q=0.8,image/png,*/*;q=0.5
Accept-Language en-us,en;q=0.5
Accept-Encoding gzip,deflate
Accept-Charset ISO-8859-1,utf-8;q=0.7,*;q=0.7
Keep-Alive 300
Proxy-Connection keep-alive
Content-Type application/x-www-form-urlencoded; charset=UTF-8
Referer http://livedemo.exadel.com/richfaces-demo/richfaces/support.jsf;jsessionid=CBD3B0656620E9A7A1AA3E5DCECAB2F4?c=support
Content-Length 125
Cookie JSESSIONID=CBD3B0656620E9A7A1AA3E5DCECAB2F4; __utmz=141944294.1201207494.8.3.utmccn=(referral)|utmcsr=jsftutorials.net|utmcct=/helpDesk/|utmcmd=referral; __utma=141944294.723785644.1197082332.1200374661.1201207494.8
Pragma no-cache
Cache-Control no-cache

This is the header not the response...

[sergeysmirnov]

"Content-Length 125" does not mean 78KB, does it?

[sergeysmirnov]

This is the header not the response...

Response is a header + context. So, if the Content-Length has 125, the header should be about 78K. Did I quote 78K above?

[stevoffm]

Hmmm...

Sergey, I think we missunderstand each other or maybe I don't understand you?!

My firebug shows an overall size of 78KB. When I click the "Response" Tab in the console it displays the WHOLE page code. Doing a similar thing in trinidad for example only produces an XHR with 800 Bytes.... When I click the "Response" Tab in this case there is NOT showing the whole page but only an xml fragment which contains the typed-in letters...

I can email you a screenshot if you want?!

Hope you understand what I mean...

[sergeysmirnov]

Yes, It sounds like we speak about different things, but I am lost to understand where is the catch.

Did you test exactly the online http://livedemo.exadel.com/richfaces-demo/richfaces/support.jsf or your own copy of richfaces-demo or your own small application or what:

What I see in the FireBug in the response. I did not count bite, but it seems for me that 125 bite guess is close to true than 78Kb:

<html xmlns="http://www.w3.org/1999/xhtml"><head><link href="/richfaces-demo/a4j_3_1_3.GAorg/richfaces

/skin.xcss/DATB/eAFjlbr0AAAC6gHS.jsf" type="text/css" rel="stylesheet" /><script src="/richfaces-demo

/a4j_3_1_3.GAorg/ajax4jsf/framework.pack.js.jsf" type="text/javascript"></script><script src="/richfaces-demo

/a4j_3_1_3.GAorg/richfaces/ui.pack.js.jsf" type="text/javascript"></script></head><body><span id="j_id213

:outtext">a</span><meta name="Ajax-Update-Ids" content="j_id213:outtext" /><span id="ajax-view-state"

><input type="hidden" name="javax.faces.ViewState" id="javax.faces.ViewState" value="_id33890" /></span

><meta id="Ajax-Response" name="Ajax-Response" content="true" /></body></html>

[stevoffm]

I test:

http://livedemo.exadel.com/richfaces-demo/richfaces/support.jsf?c=support

And it shows the complete page source code in the response tab in firebug...

strange thing... is firebug wrong?!

[sergeysmirnov]

I know that FireBug shows wrong size for network traffic when images and other resources are taken from cache, but not transfered actually. However, I never heart yet that the Ajax response tab shows something wrong.

This is a screenshot of mine:

http://a1.nupload.net/v/EPM/support.gif

[stevoffm]

Thats really weird. My firebug shows another result...

I'll recheck that with another computer...

I'll write again tomorrow, its laaaaaate.

Thanks for support so far, cya tomorrow...

[mail.micke]

Just tested this myself, and I'm seeing what Stevo is seeing.

Some pics from Firebug version 1.05

The headers:
http://www.nowupload.com/:EPq/v/

The top of the response content:
http://www.nowupload.com/:EPE/v/